sctour.predict.predict_ltsp_from_time

sctour.predict.predict_ltsp_from_time(model: sctour.train.Trainer, T: numpy.ndarray, reverse: bool = False, step_wise: bool = True, step_size: Optional[int] = None, alpha_z: float = 0.5, alpha_predz: float = 0.5, k: int = 20) → numpy.ndarray

Predict the transcriptomic latent space for query (unobserved) time intervals.

Parameters

• model – A sctour.train.Trainer for trained scTour model.

• T – A 1D numpy array containing the query time points (with values between 0 and 1). The latent space corresponding to these time points will be predicted.

• reverse – When the pseudotime returned by get_time() function for the training data was in reverse order and you used the post-inference adjustment (reverse_time() function), please set this parameter to True. (Default: False)

• step_wise – Whether to perform step-wise integration by iteratively considering only two time points when inferring the reference latent space from the training data. (Default: True)

• step_size – The step size during integration.

• alpha_z – Scaling factor for encoder-derived latent space. (Default: 0.5)

• alpha_predz – Scaling factor for ODE-solver-derived latent space. (Default: 0.5)

• k – The k nearest neighbors in the time space considered when predicting the latent representation for each query time point. (Default: 20)

Returns

Predicted latent space corresponding to the query time interval.

Return type

ndarray